44 Journal of the American Academy of Orthopaedic Surgeons Infected Total Knee Replacements Russell E. Windsor, MD, and James V. Bono, MD Next to a life- or limb-threatening complication, nothing can be more devastating than infection after total knee arthroplasty. Only through identification of risk factors and development of prophylactic regi- mens has the incidence of infection decreased. The management of this problem requires a considerable expenditure of the patient’s and the surgeon’s energy for definitive diag- nosis and treatment. Successful treatment depends on a team approach, with cooperation of the orthopaedic surgeon, the plastic sur- geon, and the infectious disease spe- cialist. Incidence of Infection The incidence of infection after total knee arthroplasty ranges from 1.1% to 12.4%. 1-5 At the Mayo Clinic, 1.2% of 3,000 primary total knee replacements developed infec- tion. 3,6 The higher rates of infection occurred after implantation of cemented linked hinges, such as the GUEPAR prosthesis (now obso- lete). 1 Patients with rheumatoid arthritis, who often are immuno- logically deficient, 7,8 are at greater risk for infection. Wilson et al 5 stud- ied 4,171 total knee arthroplasties that were performed at the Brigham and Women’s Hospital in Boston and found that 67 replace- ments became infected. The risk of infection was significantly increased in patients, particularly men, who had rheumatoid arthritis; in patients with skin ulceration; and in patients who had undergone previous knee operations. Skin infections were the most common source of infection; it is, therefore, necessary to heal skin ulcers, espe- cially those peripheral to a joint with a prosthesis. 5,9 Antibiotic pro- phylaxis is recommended until the wound is healed. In addition, infec- tion was associated with obesity, recurrent urinary tract infections, and oral corticosteroid use, although the correlation did not achieve statistical significance. Tooth extraction always causes bacteremia. 10 Although antibiotic prophylaxis in dental procedures is universally recommended, 11 it is still debated, and prospective stud- ies have shown a low risk for bacte- rial seeding around prosthetic joints. 12 Patients with chronic renal insufficiency and neoplasm requir- ing chemotherapy are at risk for infection due to chronic neutrope- nia and, in some cases, compromise of the immune system. Diabetes mellitus may pose an increased risk of infection due to the increased risk of wound-healing problems. Superficial wound necrosis may at times communicate with the deeper tissues of the knee and lead to deep infection. 13 The surgeon can influence the ulti- mate infection rate not only by tech- nique but also by selection of the prosthesis. For example, surface replacements have an overall infec- tion rate of less than 1%. In contrast, metal-on-metal constrained hinge prostheses, such as the GUEPAR prosthesis, have an infection rate that approaches 14%. Many of these infec- Dr. Windsor is Associate Professor of Surgery (Orthopaedics), Cornell University Medical Col- lege, New York, and Associate Attending Orthopaedic Surgeon, The Hospital for Special Surgery, New York. Dr. Bono is Senior Clinical Associate in Surgery, Cornell University Med- ical College, and Junior Attending Orthopaedic Surgeon, The Hospital for Special Surgery. Reprint requests: Russell E. Windsor, MD, The Hospital for Special Surgery, 535 E. 70th Street, New York, NY 10021. Copyright 1994 by the American Academy of Orthopaedic Surgeons. Abstract Deep infection is a devastating complication following total knee arthroplasty. Prompt diagnosis and definitive treatment of this complication are essential for a successful outcome. The treatment options for an infected total knee replacement include (1) antibiotic suppression alone; (2) aggressive wound debridement, drainage, and antibiotic suppression therapy; (3) resection arthroplasty; (4) arthrodesis; (5) two-stage reimplantation; and (6) amputation. Successful salvage of this complication can be accomplished only by extensive investment of surgical and infectious disease efforts in eradicating the infection. Two-stage reimplantation has been the most successful functional option and should be used whenever possi- ble to definitively eradicate the infection and ensure good function of the knee joint. J Am Acad Orthop Surg 1994;2:44-53 tions occur late, sometimes several years after implantation. The reason for this very high incidence of infec- tion is not altogether clear, but is probably related to the presence of metallic debris, which in turn causes the formation of a membranous sac containing fluid and debris around the prosthesis. 14,15 Impregnation of the bone and soft tissues with metallic fragments and the large bone-cement interface may become factors, espe- cially when the prosthesis becomes loose. Disturbingly, constrained prostheses with metal-on-plastic bearing surfaces also seem to have a higher infection rate. For example, the stabilocondylar prosthesis had a 8.3% infection rate in a small series of 36 cases. Consequently, constrained hinged prostheses with cemented intramedullary stems have become largely obsolete in this country. For the great majority of clinical situa- tions, a non- or semiconstrained sur- face total knee replacement will perform well and reduce the poten- tial for infection. Skin necrosis with secondary deep extension may lead to a deep prosthetic infection. Incisions placed at the side of the knee, for synovec- tomy or open fracture reduction and internal fixation, may predispose to skin necrosis. They are generally unsuitable for knee arthroplasty, which requires a midline, longitudi- nal incision. Although previous inci- sions should be utilized as much as possible during any knee replace- ment, sometimes it is necessary to use a separate longitudinal incision to gain exposure despite the risk of creating an island of devascularized skin between the new incision and the healed old one. A skin bridge of at least 7 cm may minimize the risk of skin necrosis. If this complication occurs, the knee should be immobi- lized until spontaneous separation of the eschar occurs. Early and aggressive attempts at debridement may lead to deep contamination that might otherwise have been avoid- able. Very large areas of necrosis, however, should be handled aggres- sively, utilizing appropriate skin grafts in consultation with a plastic surgeon. Slight wound drainage often requires no modification of the postoperative regimen. When pro- fuse wound drainage occurs, the knee should be immobilized until it stops. Antibiotics should not be given, as their administration may mask a latent deep infection. Some degree of wound drainage occurs in about 25% of the cases and may be further classified as culture-neg- ative or culture-positive. There appears to be no relationship between culture-positive wound drainage and subsequent deep infection. In the early postoperative period, a few patients have persis- tent drainage, a tense knee effu- sion, and persisting significant pain. In these patients, aggressive open debridement, evacuation of the retained hematoma, copious lavage, and reclosure should be considered. The organism most frequently found in infected total knee replace- ments is Staphylococcus aureus. Schoifet and Morrey 4 found that 58% of 31 infected total knee replacements cultured S aureus. Wil- son et al 5 observed S aureus in 42 of 67 infected replacements. Staphylo- coccal organisms were responsible for infection in the majority of patients who had concurrent skin ulcerations. Gram-negative organ- isms, such as Escherichia coli and Pseudomonas aeruginosa, have been found less frequently. A mixed polymicrobial infection is usually encountered in cases in which there is an actively draining wound through which the surface bacteria can gain entrance to the knee. In addition, patients treated with chronic antibiotic suppression may develop resistant bacterial strains. Diagnosis of Deep Infection Deep or periprosthetic infection may be either early (within 3 months of surgery) or late (more than 3 months after surgery). An early infection, provided that its course is not modified by injudicious use of antibi- otics, is usually not difficult to recog- nize. The clinical course is abnormal, with prolonged pain, swelling, inflammation, and fever. The leuko- cyte count, C-reactive protein level, and erythrocyte sedimentation rate remain elevated. Late infection is much more common than early infec- tion, and the diagnosis is usually straightforward unless antibiotics have previously been given. The usual presentation is one of acute pain and swelling in the knee of a patient with a previously satisfactory arthroplasty. Late infection usually develops from hematogenous spread of microorganisms from a distant site. Pain about a prosthesis for which a cause is not readily appar- ent should be assumed to be due to infection until proved otherwise. In a study of 52 patients with infected knee replacements treated at our institution, the preoperative work- up was evaluated for accuracy in determining infection. 16 Consider- able pain was present in 96% of the patients, 77% had swelling of the knee, 27% were febrile, and 27% had active drainage. The average erythrocyte sedimentation rate was 63 mm/h (range, 4 to 125 mm/h). The average leukocyte count was 8,300/mm 3 (range, 5,800 to 14,000/mm 3 ). Aspirated knee fluid was positive in all cases except one; in that case, no organism was cul- tured until aspiration was done at the time of the revision arthro- plasty for what was thought to be aseptic loosening. The diagnosis of an infection after total knee arthroplasty must depend Vol 2, No 1, Jan/Feb 1994 45 Russell E. Windsor, MD, and James V. Bono, MD on the results of examination of knee fluid aspirated under strict aseptic conditions. Knee radiographs are unclear in showing infection, which may be present without radiographic signs of loosening. Large complete radiolucencies usually indicate an advanced stage of infection (Fig. 1). Technetium and gallium bone scans also may not conclusively show pres- ence of infection. Cultures of wound drainage and sinus tracts, if present, often do not truly reflect the microor- ganisms found deep in the knee, since there is the likelihood of contam- ination of the fluid by other skin flora. Thus, knee aspiration is the standard of care for conclusively determining whether there is deep joint infection. The fluid aspirated from the knee is sent to the bacterio- logic laboratory for direct smear, Gram stain, and cultures with antibi- otic sensitivities for aerobic and anaerobic bacteria, acid-fast bacilli, and fungi. 16-18 If fluid cannot be easily obtained in the office, a fluoroscopi- cally assisted aspiration should be considered. If enough fluid is aspirated from the knee, a complete blood cell count and a differential white blood cell count may also give valuable infor- mation. If the former shows more than 25,000 polymorphonuclear leukocytes per cubic millimeter and the latter reveals a value greater than 75%, infection should be suspected. Fluid should also be sent for determination of glucose and protein levels. In normal synovial fluid, pro- tein levels are about a third of serum levels. Glucose values in synovial fluid are similar to those in plasma. In the presence of infection, synovial glucose values are decreased due to the presence of organisms that utilize sugar in their metabolism. Thus, low glucose and high protein values are compatible with infection. If the diagnosis is still unclear, an open biopsy by arthrotomy or arthroscopy is recommended. Frequently, patients referred from other institutions are already receiv- ing antibiotic therapy, which may suppress the infection enough to ren- der the knee aspiration fluid falsely negative. The importance of obtain- ing positive bacterial cultures preop- eratively cannot be overestimated. The cultures not only provide identification of the microorganisms but also enable the infectious disease consultant to obtain minimum serum bactericidal concentrations regularly during the course of intra- venous antibiotic therapy. If the patient has been receiving antibi- otics, they should be immediately discontinued, and serial aspirations of the knee should be done at weekly intervals until a positive culture is obtained. This method not only will provide a positive culture for the infectious disease specialist but also will increase the possibility that deep knee cultures obtained at the time of surgery will yield adequate microor- ganism growth. After antibiotics have been discontinued, it may take up to 1 month before a positive cul- ture is obtained. Patients taking antibiotics may, in fact, have positive cultures that will inaccurately reflect the bacteriologic status of the wound. In this case, other organisms may be suppressed, leading the sur- geon to erroneously think there is only a single organism present. Infections complicating primary total knee replacement should not be treated with antibiotic therapy alone. 9,15 This treatment might sup- press the symptoms of infection transiently and may be indicated only as a temporary measure if surgery is contraindicated due to medical reasons or if the patient does not accept other surgical options. Antibiotic therapy alone is unlikely to cure the infectious process. 2,19-22 Furthermore, its use can complicate the problem by selecting resistant bacterial strains. An unusual exception to this rule is the patient with a previously successful arthroplasty who presents acutely with pain, swelling of no more than 24 to 48 hours’ duration, a positive culture, and an obvious source of hematogenous bacterial contamina- tion. If aspiration demonstrates an organism that is exquisitely sensitive to antibiotic treatment, such as a Pneumococcus or Streptococcus organ- ism, antibiotics may be considered for definitive treatment. Procrastination and the prolonged use of oral antibiotics should be con- demned, particularly when infection is suspected but not confirmed by bacteriologic evidence. The end result of this course is likely to be an indolent subclinical infection and a painful prosthesis. In addition, it may make subsequent culture of the organism very difficult even after the components have been removed, so that appropriate antibiotic therapy is impossible and ultimate salvage of the arthroplasty by reimplantation becomes much less likely. 46 Journal of the American Academy of Orthopaedic Surgeons Infected Total Knee Replacements Fig. 1 Infected prosthesis 11 months post- operatively. Note radiolucency beneath tib- ial component. Treatment Options The treatment options for an infected total knee replacement include (1) antibiotic suppression alone 2,20-22 ; (2) aggressive wound debridement, drainage, and antibiotic suppression therapy 4,20,23,24 ; (3) resection arthro- plasty 25,26 ; (4) arthrodesis 22,27,28 ; (5) two- stage reimplantation 3,16-18,29 ; and (6) amputation. 2,30 Because the knee joint is relatively superficial, care of the wound is pre- eminently important. Success of any treatment option will be severely compromised by inadequate wound care or inappropriate choice of inci- sions. The original midline incision should be utilized whenever possi- ble. Provided the skin margins remain viable, the original midline incision should heal well. This inci- sion may be extended proximally and distally to improve surgical exposure of the knee joint. New inci- sions should be avoided at all costs. Well-healed medial or lateral inci- sions from operations that predate the total knee replacement should not be reopened, even if wound drainage develops in those areas. Frequently, drainage stops and the wound heals nicely after implant removal and thorough debridement. Large areas of skin necrosis or wound breakdown should be treated by rotation of a gastrocne- mius muscle pedicle graft or free vascularized muscle transfers. Antibiotic Suppression The rheumatology literature has shown that treatment of knee sepsis may be accomplished adequately by serial aspirations and antibiotic treatment. 2,20-22 However, treatment was successful in knees in which a total joint replacement was not implanted. The implant and acrylic cement act as foreign bodies that limit the ability of the immune sys- tem to adequately combat the infec- tion. However, infection is not confined to cemented total knee replacements. Wilson et al 5 found that infection developed in 2.8% of 35 uncemented total knee prosthe- ses, 1.5% of 138 hybrid total knee replacements (with an uncemented femoral component), and 1.6% of 3,998 total knee replacements with totally cemented components. These incidences were not statistically significantly different but show that infection is possible regardless of the method of implant fixation. The success of this treatment option is quite limited. 2,20-22 However, although not generally recom- mended, antibiotic suppression alone may be the only option for a patient who is a poor surgical candi- date and does not have other total joint replacements that would be at risk of becoming infected by hematogenous spread of the original infection. Only organisms with extreme sensitivity to antibiotics, such as Streptococcus species and Staphylococcus epidermidis, can be treated in this way. The disadvan- tages of this treatment include the development of resistant bacterial strains, eventual painful loosening of the prosthesis, and the risk of antibiotic toxicity due to long-term use of the medication. This method does not definitively treat the infec- tion, but rather suppresses it, and is useful only in the few patients who are so medically compromised that surgical methods would threaten their survival. Debridement With Antibiotic Suppression Therapy Vigorous wound debridement and antibiotic therapy with retention of the components has demonstrated limited success, even with the addi- tion of an ipsilateral gastrocnemius muscle flap to provide adequate soft-tissue coverage and enhance vascularity. 20,23,24 It has been found that the success is greater if infec- tion is diagnosed within 3 weeks of implantation of the original device. Schoifet and Morrey 4 specifically studied the treatment of infection after total knee arthroplasty by debridement with retention of the components. The most successful results were in seven knees in which the average time from the onset of infection to debridement was 21 days. However, the overall success rate was 23%, which reflects the fact that most of their patients had been infected for longer than 2 to 3 weeks. Borden and Gearen 17 also found that this method was somewhat more successful than more radical treatment options when the infection was diagnosed within 2 weeks of total joint implan- tation. Organisms such as Streptococcus viridans and S epidermidis may be successfully treated by this method if they demonstrate exquisite sensi- tivity to parenteral or oral antibi- otics. If this option is chosen, the patient must take antibiotics for the rest of his or her life. However, life- long antibiotic suppression poses the risk that resistant bacterial strains may develop and create breakthrough infections that are chemically difficult to treat. Patients with replacements of other joints are not usually candi- dates for debridement and suppres- sion due to the risk of hematogenous seeding of the resistant microorgan- ism strains from the site of infection to the noninfected total joint replace- ments. More radical options may become necessary if infection persists. If one thorough attempt at debridement proves unsuccessful, subsequent attempts are usually futile, and the prosthesis should be removed. Vol 2, No 1, Jan/Feb 1994 47 Russell E. Windsor, MD, and James V. Bono, MD Repeated attempts at debridement without removing the implants may compromise skin viability and may complicate definitive treatment by other surgical means. Debridement may be performed by arthroscopy or formal arthrot- omy. Formal arthrotomy allows removal of most of the scar and devi- talized tissue but may cause significant quadriceps weakness in the postoperative period due to the incision through the extensor mech- anism. The surgeon should remove all synovium and scar tissue and clear the medial and lateral gutters of debris. It may be necessary to fully expose the knee replacement in order to properly debride the poste- rior joint capsule. Arthroscopic intervention may accomplish the same goals; how- ever, multiple (up to six) portals may be required. The procedure is gener- ally longer than arthrotomy because of the slower extraction of tissue by rotary suction blades. Infections that create significant scarring may ren- der arthroscopy impossible as a treatment option. Regardless of the surgical method used, a thorough debridement is done. Frozen tissue sections, Gram stains, cultures of the tissue, and the macroscopic appearance of the wound should provide diagnostic information. After debridement, the wound is closed over suction drains, which should remain in place for 36 to 48 hours. Using ingress and egress tubes with continuous irrigation is no longer recommended, as there is a significant risk of fluid extravasa- tion as well as a risk of exogenous superinfection due to communica- tion of the deep anatomic structures with the skin. Under no circum- stances should the wound be left open to close by secondary forma- tion of granulation tissue. The wound is inspected after 2 weeks and is reaspirated under strict aseptic conditions. If the wound is benign and the cultures are negative, antibiotic therapy is continued for a further 4 weeks. When this is not the case, reoperation with removal of the prosthetic components and all cement is performed. This decision should be made quickly before fur- ther compromise of the underlying tissues develops. Resection Arthroplasty The two previous treatment options preserve the total joint replacement. Due to their limited success, more radical surgical options are usually required to eradicate the deep infec- tion. Resection arthroplasty involves the complete removal of all compo- nents of the knee replacement, acrylic cement, scar tissue, and syn- ovium 25,26 (Fig. 2). This option as a definitive proce- dure is generally reserved for med- ically fragile patients who cannot tolerate another major operation. It may also serve as an intermediate step for the patient who has reserva- tions concerning arthrodesis. Falahee et al 25 reported on 28 knees that under- went resection arthroplasty because of infection after total knee arthro- plasty. Eleven patients had multiartic- ular rheumatoid arthritis, 14 had osteoarthritis, and one patient had multiarticular neuropathic arthropa- thy. Six patients with monarticular osteoarthritis considered resection arthroplasty unacceptable and subse- quently underwent successful sec- ondary arthrodesis. In three patients, spontaneous bone fusion developed after the resection with the knee in good position. The patients who had had the most severe disability before total knee arthroplasty were the most likely to be satisfied with the func- tional results of resection arthro- plasty. Conversely, the patients who had had the least severe disability were more likely to find the results of resection arthroplasty unacceptable. Fifteen patients were able to walk independently without assistance. Five of those patients were able to stand and walk without external limb support. The other 10 patients used either a knee-ankle-foot orthosis or a universal knee splint. All 15 patients, however, required either a cane or a walker and remained either moder- ately or severely restricted in their overall walking capacity. Resection arthroplasty is very useful for the severely disabled per- son with a sedentary lifestyle. The procedure is least suitable for patients who had a relatively minor disability before their original total joint replacement. Those patients will obtain more tolerable function from arthrodesis or reimplantation of a total knee replacement, depend- ing on the sensitivity of the infective organism and the adequacy of the antibiotic treatment. The advantage of resection arthroplasty is that some motion is preserved to allow sitting and to 48 Journal of the American Academy of Orthopaedic Surgeons Infected Total Knee Replacements Fig. 2 Resection arthroplasty in a med- ically fragile patient following sepsis. facilitate transferring into and out of automobiles and aircraft. The disad- vantage is the possibility of persis- tent pain and instability on walking. Arthrodesis Arthrodesis may be the only option for treating the infected total knee replacement when other forms of treatment are contraindicated. 28 Suc- cessful arthrodesis depends mainly on technique and the availability of adequate bone to accomplish fusion. The success of arthrodesis can be as low as 50% when it is used to salvage an infected hinge prosthesis 1 (Fig. 3). Significant bone loss is often associ- ated with removal of these implants, making the remaining bone stock inadequate to attain successful fusion. Successful arthrodesis has been accomplished in as many as 90% of the cases in which surface replacements were utilized and bone stock was well preserved. 22,28 The indications for arthrodesis are (1) complete destruction of the extensor mechanism by infection, rendering the patient incapable of actively extending the knee; (2) a resistant bacterial infection that requires high toxic doses of antibi- otic therapy to reach adequate bacte- ricidal levels; (3) a knee with inadequate bone stock for placement of a new total knee prosthesis; (4) a knee with inadequate soft-tissue coverage and multiple incisions that may compromise future wound healing; and (5) a young patient in whom the likelihood of subsequent infection or revision is great. Arthrodesis may be accom- plished by different techniques. Adequate bacteriologic control of the wound should be obtained beforehand. We do not recommend performing arthrodesis at the time of the original debridement, as the risk of persistent infection is high in the setting of active wound sepsis when metallic implants are needed to accomplish fusion. Therefore, we perform arthrodesis in a staged manner. However, some advocate immediate arthrodesis. 28 Our protocol is as follows: The first stage involves complete removal of the prosthesis and cement, along with complete wound debridement. The second stage involves a 4- to 6-week course of intravenous antibiotic therapy, maintaining a minimum bactericidal concentration of 1:8. Arthrodesis is then performed as the final stage. External fixation and intra- medullary rod fixation are two meth- ods of arthrodesis particularly applicable to this clinical situation. External fixation is particularly appropriate in patients who have an ipsilateral total hip replacement above the affected knee joint and in patients with an especially virulent microorganism. In the second method, a curved intramedullary rod is placed through the knee joint from the greater trochanter of the femur down to the distal aspect of the tibia just above the malleoli. The widest-diam- eter rod that can fit in the tibial intramedullary canal should be used. Although Puranen et al 28 believe that no secondary bone grafting is needed with this technique, ade- quate bone may be obtained from the anterior tibial flare or the patella to aid fusion. Care should be taken to provide adequate bone contact between the femur and the tibia. Postoperatively, the patient may begin ambulating immediately with- out the need for external supports. Puranen et al reported the success of intramedullary arthrodesis in 33 patients. Fifteen patients had failed total knee replacements, eight of which were infected. In the 33 cases, four nails broke, three at the line of fusion and one in the area of an infected supracondylar pseudarthro- sis. Therefore, protected weight- bearing should be maintained until fusion is proved radiographically. The advantages of arthrodesis as treatment for an infected total knee replacement are that it is a definitive treatment for the infection with little chance of recurrence and that it promises reasonably good long- term function without the risk of future mechanical failure. The dis- advantages of arthrodesis are inabil- ity to bend the knee; difficulty in transferring from a car or sitting in a small space, such as an airplane; and the large increase in the energy required to walk with a stiff knee, which may be a particular problem for patients with cardiovascular and pulmonary problems. Two-Stage Reimplantation The most successful functional results for the treatment of late infec- tion of a total knee replacement are obtained by a technique of two-stage reimplantation of a new total knee replacement, 3,16-18,29,31 with success Vol 2, No 1, Jan/Feb 1994 49 Russell E. Windsor, MD, and James V. Bono, MD Fig. 3 Hinged prosthesis with severe periprosthetic bone loss. rates averaging 90%. This method represents the procedure of choice to definitively eradicate the infection and preserve knee function. Adequate preoperative planning is necessary and the availability of special instruments is recom- mended. Removal of the prosthetic components and acrylic cement can prove difficult, particularly if the septic process is of recent onset. In this case, the prosthetic components will most likely be well fixed, and removal of the tight interdigitation between bone and cement demands meticulous technique in order to pre- vent unnecessary loss of bone stock. The removal of hinged total knee replacements with intramedullary stems in the femur and tibia can also prove difficult. For these cases, spe- cial cement osteotomes and a high- speed cement drill are helpful. Surgical Protocol The protocol involves three stages. 16,19 The patient must be in good general medical health to with- stand the rigors of all the stages. The first stage of the protocol involves complete debridement of all infected tissues, along with removal of the implants and all cement. All scarred, inflamed, and devitalized tissues should be thor- oughly excised, leaving viable, healthy, well-vascularized tissues. Primary wound closure can usually be performed over closed suction tubes, which are removed after 24 to 48 hours. The knee is immobilized in a bulky Robert Jones dressing with plaster splints. During this initial debridement, a central intravenous access catheter (e.g., a Broviac or Hickman catheter) is introduced into the internal or external jugular vein to facilitate intravenous administra- tion of antibiotics. The dressing is changed after a few weeks to a hinged brace, with the knee in full extension or 5 degrees of flexion. Skeletal traction is not advised, and the bone ends are left in contact to reduce dead space. Some surgeons recommend the use of antibiotic-impregnated spacers to preserve the joint space and to facil- itate wound exposure during the reimplantation stage of the proto- col. 31,32 The spacer block does repre- sent a foreign body in the knee joint, however, and would remain in place in the event of a medical complica- tion. The second stage involves a 6- week course of intravenous antibi- otic therapy, based on the results of culture and sensitivity studies. Antibiotics are chosen to yield high bactericidal effect with low toxicity. The infectious disease consultant follows the minimum bactericidal concentrations and weekly exam- ines the efficacy of the antibiotic blood concentration against the bac- teria available on the culture media. A minimum bactericidal concentra- tion of 1:8 must be maintained for the 6-week period. The time between the first and last stages of the proto- col may be prolonged if this mini- mum bactericidal concentration is not achieved. After 6 weeks, if the wound is completely benign and the patient has had an uneventful postoperative period, another total knee prosthesis may be inserted. 16,33 However, if the wound still shows signs of inflamma- tion, reimplantation is postponed; a long cylinder cast is applied, and the patient starts ambulation and is dis- charged home as soon as he is inde- pendent. Further inspection of the wound is made a month later. If it is benign, reimplantation is considered. Alter- natively, the knee joint fluid may be serially aspirated after discharge to determine whether there is persis- tence of infection. The last stage of the protocol involves implantation of a new total knee replacement. Frequently, a modular prosthesis is used, which enables the surgeon to reconstruct any bone loss by adding metal wedges to the tibial component and distal and/or posterior augmenta- tion to the femoral component. Frozen tissue sections and Gram stains are obtained at the time of surgery to assess tissue inflamma- tion. The macroscopic appearance of the wound should be completely benign; all scarred and devitalized tissue is excised, leaving only viable, well-vascularized, healthy tissues. Exposure can sometimes be difficult after prolonged immobilization; there is a danger of avulsing the tib- ial tubercle while attempting to mobilize the patella and flex the knee. If this event seems likely, either a quadriceps snip or a turn- down is used. Preoperative planning is essential in order to have adequate prosthetic components available. A special cus- tom-designed prosthesis is occasion- ally necessary. In most cases, proper alignment can be reestablished with adequate tissue tension and the use of press-fitted fluted intramedullary rods. The proximal end of the tibial component and the distal end of the femoral prosthesis are cemented. Some surgeons cement the prosthe- sis completely. Excision of the patella has proved helpful in cases in which the skin closure was too tight. If the patella has insufficient bone stock to accept a prosthesis, it may be left unresurfaced. Normally, recon- struction can be achieved using standard designs that provide a substitution for the posterior cruci- ate ligament; in some cases, designs that preserve the posterior cruciate ligament are used. In our experience, revision arthroplasty in a wound without an acrylic spacer block may be accom- plished with only slightly greater difficulty than in one with a spacer. Other surgeons believe the use of acrylic spacers makes later surgery 50 Journal of the American Academy of Orthopaedic Surgeons Infected Total Knee Replacements significantly less difficult. They also feel that the extension of the spacer into the suprapatellar pouch reduces scarring. The use of constrained compo- nents is often unavoidable. When this is the case, a constrained condylar knee prosthesis (e.g., the Total Condylar III) is selected; such a device has intramedullary stems on both components and restricts varus/val- gus, anteroposterior, and rotary motions by means of a centrally posi- tioned peg. Intramedullary stems are fitted in a modular fashion to the femoral or tibial prosthesis and are press-fitted into the intramedullary canal. While a stemmed component in the tibia, femur, or both is required because of bone deficiency, constraint at the prosthetic surfaces is not auto- matically required unless there is uncontrollable ligamentous insta- bility. The use of antibiotic-impregnated cement has been recommended for reimplantation after infection. 31,34 However, the effectiveness of this technique in preventing reinfection has not been statistically proved. Postoperative Management Postoperative management after reimplantation is the same as that used after a primary arthroplasty unless a quadriceps turndown was done to facilitate exposure. In this event the knee is immobilized with plaster splints for 3 weeks before motion is begun. On the assumption that the knee is sterile at the time of reimplantation, there is no need for prolonged antibiotic therapy. Peri- operative antibiotics are adminis- tered for 4 days until the final operative culture readings are obtained, after which time no addi- tional antibiotics are given. In most cases the antibiotic that is used for the 6-week course is continued through the postoperative period, and additional antibiotics may be given to provide broader antimicro- bial coverage during the reimplanta- tion procedure. Results Since 1977, it has been the policy of the knee service at our institution to implant a new prosthesis when- ever possible. Between January 1977 and December 1985, 48 patients in whom 52 total knee arthroplasties were complicated by infection were admitted for treatment. 16 Four patients had bilateral infections. Thirteen patients (14 knees) were excluded from study for a variety of reasons. The reasons for exclusion can be reduced to a few: (1) patient preference, (2) antibiotic toxicity risk, (3) medical infirmity, and (4) inadequate skin viability and exten- sor mechanism function. With an average follow-up time of 4 years (range, 2.5 to 10 years), 37 of 38 knees were successfully treated without recurrence of the original infection (97% success rate). In the 37 patients who underwent the proce- dure, there were 11 excellent, 13 good, 6 fair, and 7 poor results, based on the Hospital for Special Surgery Score. The reasons for the poor results were reinfection by a different organism and compromise of exten- sor mechanism function with persis- tent pain. Overall function was well maintained in the group, with a range of motion averaging 95 degrees (range, 80 to 120 degrees). Twenty-three patients complained of some pain when walking, 15 patients had mild pain, 6 patients had moder- ate pain, and 2 patients had severe pain. All patients underwent reim- plantation of a cemented prosthesis, which frequently had press-fitted intramedullary stems. Although Insall et al 18 cautioned against using this protocol for reim- plantation in the presence of Gram- negative infection, it has been more recently shown by Windsor et al 16 that it is feasible to perform this pro- tocol when certain sensitive Gram- negative infections are present. Escherichia coli and P aeruginosa infections have been successfully treated with this protocol, and the presence of newer nontoxic bacteri- cidal agents has made it possible to eradicate these Gram-negative infec- tions. Other authors have tried to accomplish successful reimplanta- tion by utilizing shorter periods of intravenous antibiotic therapy. However, Rand and Bryan 3 found a 2-week course of therapy totally unacceptable in definitively eradi- cating the infection. Borden and Gearen 17 studied a small number of infections that had been treated with a 4-week course of intravenous antibiotics. The overall results fell between the poor results reported by Rand and Bryan 3 with a 2-week course and the excellent results obtained by Windsor et al 16 and Insall et al. 18 Therefore, use of a 6- week course of intravenous antibi- otic administration can ensure success in the majority of patients with an infected knee prosthesis, with a 97% overall success rate in eradicating the original infec- tion. 16,18,29,31 It is our opinion that the most suc- cessful method of treating infection of a total knee replacement with definitive eradication of the bacteri- ologic organism is by the two-stage reimplantation procedure. This pro- tocol requires meticulous attention to detail by the surgical and infec- tious disease staff, but a predictable outcome can be ensured in most cases. Amputation Amputation may be the final salvage procedure for severe infections that are associated with large-bone loss and compromised antibiotic treat- ment. 30 This procedure was required most frequently in infected knee Vol 2, No 1, Jan/Feb 1994 51 Russell E. Windsor, MD, and James V. Bono, MD 52 Journal of the American Academy of Orthopaedic Surgeons Infected Total Knee Replacements References 1. Deburge A, GUEPAR Group: Guepar hinge prosthesis: Complications and results with two years’ follow-up. Clin Orthop 1976;120:47-53. 2. Grogan TJ, Dorey F, Rollins J, et al: Deep sepsis following total knee arthroplasty: Ten-year experience at the University of California at Los Angeles Medical Cen- ter. J Bone Joint Surg Am 1986;68:226-234. 3. Rand JA, Bryan RS: Reimplantation for the salvage of an infected total knee arthroplasty. J Bone Joint Surg Am 1983;65:1081-1086. 4. Schoifet SD, Morrey BF: Treatment of infection after total knee arthroplasty by débridement with retention of the com- ponents. J Bone Joint Surg Am 1990;72: 1383-1390. 5. Wilson MG, Kelley K, Thornhill TS: Infection as a complication of total knee- replacement arthroplasty: Risk factors and treatment in sixty-seven cases. J Bone Joint Surg Am 1990;72:878-883. 6. Rand JA, Bryan RS, Morrey BF, et al: Management of infected total knee arthroplasty. Clin Orthop 1986;205:75-85. 7. Garner RW, Mowat AG, Hazleman BL: Wound healing after operations on patients with rheumatoid arthritis. J Bone Joint Surg Br 1973;55:134-144. 8. Infection in rheumatoid disease [ed- itorial]. BMJ 1972;2:549-550. 9. Thomas BJ, Moreland JR, Amstutz HC: Infection after total joint arthroplasty from distal extremity sepsis. Clin Orthop 1983;181:121-125. 10. Lindqvist C, Slätis P: Dental bacteremia: A neglected cause of arthroplasty infec- tions? Three hip cases. Acta Orthop Scand 1985;56:506-508. 11. Nelson JP, Fitzgerald RH Jr, Jaspers MT, et al: Prophylactic antimicrobial cover- age in arthroplasty patients [editorial]. J Bone Joint Surg Am 1990;72:1. 12. Ainscow DAP, Denham RA: The risk of haematogenous infection in total joint replacements. J Bone Joint Surg Br 1984;66:580-582. 13. England SP, Stern SH, Insall JN, et al: Total knee arthroplasty in diabetes mel- litus. Clin Orthop 1990;260:130-134. 14. Rae T: A study on the effects of particu- late metals of orthopaedic interest on murine macrophages in vitro. J Bone Joint Surg Br 1975;57:444-450. 15. Schurman DJ, Johnson BL Jr, Amstutz HC: Knee joint infections with Staphylo- coccus aureus and Micrococcus species: Influence of antibiotics, metal debris, bacteremia, blood, and steroids in a rab- bit model. J Bone Joint Surg Am 1975;57:40-49. 16. Windsor RE, Insall JN, Urs WK, et al: Two-stage reimplantation for the sal- vage of total knee arthroplasty compli- cated by infection: Further follow-up and refinement of indications. J Bone Joint Surg Am 1990;72:272-278. 17. Borden LS, Gearen PF: Infected total knee arthroplasty: A protocol for man- agement. J Arthroplasty 1987;2:27-36. 18. Insall JN, Thompson FM, Brause BD: Two-stage reimplantation for the sal- vage of infected total knee arthroplasty. J Bone Joint Surg Am 1983;65:1087-1098. 19. Insall JN: Infection in total knee arthro- plasty. Instr Course Lect 1982;31:42-48. 20. Johnson DP, Bannister GC: The outcome of infected arthroplasty of the knee. J Bone Joint Surg Br 1986;68:289-291. 21. Marsh PK, Cotler JM: Management of an anaerobic infection in a prosthetic knee with long-term antibiotics alone: A case report. Clin Orthop 1981;155: 133-135. 22. Woods GW, Lionberger DR, Tullos HS: Failed total knee arthroplasty: Revision and arthrodesis for infection and nonin- fectious complications. Clin Orthop 1983;173:184-190. 23. Peled IJ, Frankl U, Wexler MR: Salvage of exposed knee prosthesis by gastroc- nemius myocutaneous flap coverage. Orthopedics 1983;6:1320-1322. 24. Sanders R, O’Neill T: The gastrocne- mius myocutaneous flap used as a cover for the exposed knee prosthesis. J Bone Joint Surg Br 1981;63:383-386. 25. Falahee MH, Matthews LS, Kaufer H: Resection arthroplasty as a salvage pro- cedure for a knee with infection after a total arthroplasty. J Bone Joint Surg Am 1987;69:1013-1021. 26. Lettin AWF, Neil MJ, Citron ND, et al: Excision arthroplasty for infected con- strained total knee replacements. J Bone Joint Surg Br 1990;72:220-224. 27. Hagemann WF, Woods GW, Tullos HS: replacements with cemented, stemmed hinges, which for the most part have become obsolete. The remaining shell of bone was fre- quently inadequate for subsequent arthrodesis or reimplantation, mak- ing the limb essentially flail. Ampu- tation may be the only option in patients with mixed infection for whom antibiotic treatment has proved inadequate or in whom there is such massive tissue destruction that knee function is unsalvageable. This frequently occurs with mixed infections in which multiple ab- scesses and sinus tracts are present and significant destruction of the surrounding soft-tissue sleeve and muscle occurs. If successful treat- ment cannot be accomplished in any other way, a successful above-knee amputation may provide the best function for patients who otherwise would have a functionless knee joint and distal extremity. Summary Successful treatment of the infected total knee replacement represents the most difficult form of revision surgery. A two-stage reimplanta- tion procedure for treating this con- dition has been the most successful functional option and should be uti- lized whenever possible to defi- nitively eradicate the infection and ensure good function of the knee joint. It is the procedure of choice for younger healthy patients who can medically tolerate the protocol. Other options may be necessary, depending on the medical condition of the patient. Initial debridement with antibiotic suppression should be considered only in infected total knee replacements during the first 2 to 3 weeks after the operation. Antibiotic suppression is limited to the small minority of patients medically unable to tolerate fur- ther surgery and to those few patients with acute hematogenous infection of 24 to 48 hours’ dura- tion in whom the infecting organ- isms are exquisitely sensitive to antibiotics. Vol 2, No 1, Jan/Feb 1994 53 Russell E. Windsor, MD, and James V. Bono, MD Arthrodesis in failed total knee replace- ment. J Bone Joint Surg Am 1978;60: 790-794. 28. Puranen J, Kortelainen P, Jalovaara P: Arthrodesis of the knee with intra- medullary nail fixation. J Bone Joint Surg Am 1990;72:433-442. 29. Rosenberg AG, Haas B, Barden R, et al: Salvage of infected total knee arthro- plasty. Clin Orthop 1988;226:29-33. 30. Pring DJ, Marks L, Angel JC: Mobility after amputation for failed knee replace- ment. J Bone Joint Surg Br 1988;70: 770-771. 31. Wilde AH, Ruth JT: Two-stage reim- plantation in infected total knee arthro- plasty. Clin Orthop 1988;236:23-35. 32. Carlsson AS, Josefsson G, Lindberg L: Revision with gentamicin-impregnated cement for deep infections in total hip arthroplasties. J Bone Joint Surg Am 1978;60:1059-1064. 33. Freeman MAR, Sudlow RA, Casewell MW, et al: The management of infected total knee replacements. J Bone Joint Surg Br 1985;67:764-768. 34. Marks KE, Nelson CL, Lautenschlager EP: Antibiotic-impregnated acrylic bone cement. J Bone Joint Surg Am 1976;58: 358-364.